DATA
Bits
Everything you see on your computer screen is a digital representation. All
computer data is expressed in bits, which are either 0's or 1's. "We might
call these values 0V and 3.5V, or on or off, or true or false, or yin or yang,
but we conventionally denote them as 0 and 1."
Bits,
Bytes and Displays
Bits
are grouped into larger units called bytes. There are 8 bits to one byte. One
byte may have the binary number 01100001 (8 bits) which is equal to the value
97. This number can be translated into a color, a letter, a brightness or whatever
the software and hardware and their interface determines it to be. In the ASCII
character set it assigns the bit pattern 01100001 (value 97), the lower case
letter "a". When you press the key "A" on your keyboard,
it sends an electrical pattern (0's and 1's) or the value 97 to your computer
and software to display "a".
In terms of color depth, or the complexity of color in an image, the number of bits determines the color displayed. A single bit is two colors; 4 bits produces 16 colors; 8 bit color depth means that each pixel could one of 256 colors. Each 256 colors has a number or value associated with it; for instance, 0 = Black and 255 = White. You have probably noticed this if you use Photoshop and color sliders. 64 color bit depth means each pixel could be one of a million plus colors, giving you a wide range of hue values. However, the more bits, the larger the file size--for 64 bit color (millions of colors), each pixel requires 64 bits, rather than 8 bits for 8 bit color. The more color variation, the larger the file size. Before broadband was developed or commonly used, many web designers used 2 or 4 bit color which produced interesting graphics. This is one of the ways that limitations of the web can be used creatively. For a good description of various bit depths and the difference between using gif or jpeg, follow this link.
Data
Rates
In streaming media, we talk about data rates, the amount of data which moves
through a connection, in terms of bits or kilobits or kbits or kb. This is not
to be confused with KB or Kilobytes or KBytes or KB; this expression is more
often used with multimedia files in which data transfer rates can be much higher--in
the megabytes rather than bits. For streaming purposes, we almost always speak
in "bits".
However,
here are some numbers to remember:
8 bits = 1 byte
1000 bits = 1 kilobit (kbit) (a 56k Modem is 56kbits/sec; or 56, 000 bits a
sec--theoretically)
1024 bytes = 1 Kilobyte (KByte or 1K)
1000 KB = 1 Megabyte (MB)
1000 MB = 1 Gigabyte (1 GB)
Digital Video's (DV) transfer rate is 3.6 MBytes per second. (30,000 kbits a
second)
1 minute of uncompressed DV = 200 MB
5 minutes of uncompressed DV = 1 GB
Data Rates and Network Connections
If
you look in your reader, at Cleaner 6 Manual Pages, a. Data Rates, you will
see common bit rates for different network connections. These are the network
bandwidths, or how much data can pass through the connnection from the server
to your or anyone's computer. Think of this as pipes though which water flows.
The larger the pipe the more water (more data) and the least resistence (faster,
quieter). If you look at the 56.6k modem connection, it is about 32 kbits per
second (kbps). (or
4 KBytes. Do the math--8 bits to the byte, so divide 8 into 32 to get 4KB. However,
it is always better to use kbits when discussing network connections.)
Below are some connection speeds and what you might expect. Actual throughput
is of course effected by complexity of video image, size of image, type of encoding,
length of video and number of people connected to stream and so on. Different
manuals will give you different data rates. Many providers oversell the capacity
for connectivity to their server, sometimes up to 10 to 1. A 300 kbps DSL connection
may have the actual throughput of a 28.8 modem in real life.
Network Connection | kbit/sec | audio quality | video quality |
28.8K modem | 20kbps | mono FM | low motion content; small screen 160x120; 1-10 fps |
56.6Kmodem | 32kbps | FM, possibly stereo | low motion content; small screen 160x120; 5-12 fps |
ISDN | 64 - 96kbps | as good as CD | good low motion content, ok high motion; screen size determines quality; 160x120 good; 240x180 and 320x240 compromised image quality; 10-30 fps |
Cable Modem | 128-300 kbps | CD transparent | very good quality (VHS and a better) at 300kbps; 320x340 screen good |
LAN | 160-800 kbps | CD transparent | very good quality (VHS and better), 320x240 and higher; 15-30 fps |
Want to know what
kind of real connectivity you have at home? Try downloading a large uncompressed
file, measured usually in Kbytes (such as one of the players we are using).
Multiply this by 8 (8 bits = 1 byte), then divide this by number of seconds
it took to download. This would be your current throughput. Try this at different
times of the day.
Compression Strategies
In general
we can reduce file size in three ways:
1. Reduce the data-rate by increasing compression
2. Reduce the physical size or pixle dimensions of the video
3. Reduce the number of frames per second
As for #1, compression can happen either spatially or temporally, or both. Different
codecs apply different compression schemes, some relying more heavily on spatial
compression (compression within the frame) and some on temporal
compression (compression across or between frames). The trick is to
experiment with all three of these factors, until you find a combination which
suits your need for your particular video. Spatial compression is often expressed
in export options (in FCP) as "quality", where as temporal is expressed
through "keyframes".
For more information on compression basics, visit this website.